Efficacy and Safety of Continuous Glucose Monitoring and Intermittently Scanned Continuous Glucose Monitoring in Patients With Type 2 Diabetes: A Systematic Review and Meta-analysis of Interventional Evidence

In managing type 2 diabetes (T2D), the goals of treatment are to prevent or delay complications, and this requires optimal control of glycemia and cardiovascular risk factors such as lipids and blood pressure (1,2). Management involves various approaches, including lifestyle modifications, pharmacotherapy, and self-monitoring of blood glucose (SMBG), which requires several measurements of blood glucose taken during the course of the day. Traditionally, SMBG has been the cornerstone of diabetes management (3); however, it has limitations given the limited data provided and inconvenience of several finger pricks each day (4). With advancements in technology minimally invasive devices have been introduced such as continuous glucose monitoring (CGM) and intermittently scanned CGM (isCGM) systems as potential tools for managing T2D more effectively. CGM offers a continuous measurement of interstitial glucose levels in subcutaneous tissues, providing individuals with comprehensive glucose data on their glycemic patterns, trends, and fluctuations (5). isCGM provides the same type of glucose data as CGM but requires the user to purposely scan the sensor to obtain information (6). While CGM and isCGM have significant advantages over SMBG and have shown considerable promise in improving glycemic control and other clinical outcomes in individuals with type 1 diabetes (T1D) (7–10), their efficacy and impact on T2D management have been inconsistent. CGM includes real-time CGM and retrospective (professional) CGM; real-time CGM provides blood glucose profile in real time, whereas retrospective CGM is used to retrospectively examine lifestyle problems and pharmacotherapy adjustment following observation of the blood glucose profile over several days (11). Though a number of studies performed in patients with T1D suggest that real-time CGM improves glucose levels better than retrospective CGM (12,13), the differential efficacy of these CGM types compared with usual care or SMBG in patients with T2D is uncertain. Furthermore, whether the efficacy of CGM or isCGM varies in the more heterogeneous patients with T2D who can be on different treatment modalities is unclear. Finally, the minimum time of use required for CGM or isCGM to be effective in patients with T2D remains unknown. Given the uncertainty of evidence, it would be valuable to collectively assess available studies to provide a comprehensive understanding of the role of these two glucose monitoring systems (CGM and isCGM) in the management of T2D. We conducted a systematic review and meta-analysis to evaluate the benefits and harms of CGM and isCGM compared with usual care or SMBG in patients with T2D, while also attempting subgroup analyses according to relevant clinical characteristics such as device type, length of use, and diabetes treatment.

The predefined protocol for this systematic review and meta-analysis was prospectively registered in the International prospective register of systematic reviews (PROSPERO) (CRD42023444691). Our methodology and reporting strictly adhered to the guidelines outlined in Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) (Supplementary Appendix 1). We conducted a systematic search without language restrictions, using MEDLINE, Embase, and the Cochrane Library databases, from inception to 3 August 2023. The MEDLINE search strategy can be found in Supplementary Appendix 2, and this was subsequently adapted for the other databases. To facilitate the screening process, we used Rayyan, an online bibliographic tool (https://rayyan.qcri.org) (14). One author (S.K.K.) performed the initial screening, and subsequently two authors (S.S. and S.K.K.) conducted a thorough evaluation of the full texts of potentially eligible articles. In cases where disagreements arose regarding the eligibility of an article, a discussion took place, and consensus was reached through the involvement of a third author.

We included studies in our analyses if they met the following criteria: randomized controlled trials (RCTs) with comparison of at least two of the following interventions for a duration of at least 8 weeks: CGM in real-time or retrospective mode, intermittent short-term use of CGM, long-term use of CGM and isCGM, and usual care/SMBG in patients with T2D. We required that studies included reporting of data on glycemic measures and other relevant outcomes.

In our data extraction process, we used a standardized predesigned data collection form (15,16). The following information was extracted: the first author and year of publication, study design characteristics, baseline characteristics, and characteristics of the intervention and comparator, as well as the outcomes of interest. To ensure accuracy and consistency, one experienced reviewer (S.K.K.) performed the initial data extraction. Subsequently, a second experienced reviewer (S.S.) independently cross-checked the extracted data.

The prespecified primary outcome was change in HbA_1c levels. Secondary outcomes included 1) other glycemic measures including fasting plasma glucose (FPG), 2) CGM metrics (e.g., blood glucose time in range [TIR], time below range [TBR], time above range), 3) body composition measures (weight, BMI), 4) metabolic outcomes (blood pressure, lipids), 5) medication effect score, 6) safety events, and 7) psychological outcomes (measures of satisfaction, distress, and quality of life).

To assess the risk of bias in the RCTs included in our analysis, we applied the Cochrane Collaboration’s risk-of-bias tool (17). Each study was categorized as having a low, unclear, or high risk of bias for each domain. Additionally, we used the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach, implemented through the GRADEpro tool (https://gdt.gradepro.org), to assess the certainty of the evidence for each outcome (18). The certainty of evidence was rated at four distinct levels: high, moderate, low, and very low.

Summary measures of effect were presented as mean differences (MDs) with 95% CIs for continuous outcomes and relative risks (RRs) with 95% CIs for binary outcomes. For continuous data measured on different scales, we used the standardized MD (SMD) with 95% CIs. Due to the varying treatment periods reported across studies, we used effect estimates (RRs, MDs, and SMDs) from the longest treatment period for each outcome in the main analysis. For the outcome of change in HbA_1c, which included data from ≥10 studies, we also estimated pooled effect estimates at specific time points: 12 weeks (±4), 24 weeks (±4), and ≥30 weeks. These time points were chosen based on the distribution of time points reported in the eligible studies and to maintain consistency with previous related reviews (19,20). To account for between-study heterogeneity, we used random-effects models for pooling the measures of effect. We assessed the extent of statistical heterogeneity across studies using standard χ² tests and the I² statistic. We explored prespecified study-level characteristics as potential sources of heterogeneity using stratified analysis and random-effects meta-regression. For the outcome of change in HbA_1c, which was based on pooled analysis of ≥10 studies, we also assessed for small study effects, including publication bias, using formal tests such as Begg funnel plots (21) and Egger regression symmetry test (22). All statistical analyses were performed with Stata, version MP 17 (StataCorp, College Station, TX).

Figure 1 outlines our study selection process. We initially identified 1,199 potential citations from various sources. After rigorous screening, 29 articles from 26 distinct RCTs met our inclusion criteria (Supplementary Appendix 3).

The publication dates of included articles ranged from 2008 to 2023. Relevant baseline characteristics of the individual RCTs are summarized in Table 1. The comparison between CGM and usual care/SMBG involved 17 unique RCTs comprising 1,146 patients with T2D (632 assigned to CGM and 514 assigned to usual care/SMBG). That between isCGM and usual care/SMBG involved nine unique RCTs comprising 1,637 patients with T2D (871 assigned to isCGM and 766 assigned to usual care/SMBG). Patient ages ranged from 53 to 70 years with a weighted mean age of 60.1 years. Duration of T2D ranged from 5.6 to 21.8 years. Patients were on various diabetes treatment regimens such as oral antidiabetes drugs without insulin, oral antidiabetes drugs and/or insulin, and insulin. Baseline HbA_1c levels ranged from 6.9 to 9.9%. For studies of CGM, CGM was either real-time or retrospective and the duration of intervention ranged from 8.0 to 34.8 weeks with a weighted mean intervention duration of 20.3 weeks. The respective figures for isCGM were 10–52 weeks and 32.4 weeks. With use of the Cochrane risk-of-bias tool, all 26 trials demonstrated a low risk of bias in incomplete outcome data and selective reporting. All but four trials had a low risk of bias in random sequence generation (Appendix 4).

Use of CGM, versus usual care/SMBG, reduced levels of HbA_1c (n = 14 studies) in analyses based on the longest treatment period for each study: MD −0.19% (95% CI −0.34, −0.04) (Fig. 2A). There were no differences in HbA_1c levels in comparing CGM with usual care/SMBG at the specific time points of 12 (n = 6 studies), 24 (n = 5 studies), and ≥30 weeks (n = 3 studies): MD −0.12% (−0.31, 0.08), −0.23% (−0.59, 0.13), and −0.19% (−0.54, 0.16), respectively (Fig. 2B).

isCGM, versus usual care/SMBG, reduced levels of HbA_1c (n = 8 studies) in analyses based on the longest treatment period for each study: MD −0.31% (95% CI −0.46, −0.17) (Fig. 3A). isCGM as compared with usual care/SMBG reduced levels of HbA_1c at the specific time points of 12 (n = 4 studies) and 30 weeks (n = 3 studies) but not 24 weeks (n = 1 study): MD −0.38% (−0.59, −0.17), −0.35% (−0.54, −0.16), and 0.03% (−0.26, 0.32), respectively (Fig. 3B).

The effect of CGM versus usual care/SMBG on HbA_1c levels did not vary significantly by age, T2D treatment, T2D duration, baseline HbA_1c level, CGM type (real time vs. retrospective), or CGM use (P value for meta-regression for all >0.05) (Supplementary Appendix 5). There were no differences in other HbA_1c measures such as HbA_1c level <7.0% (n = 3 studies), HbA_1c level <7.5% (n = 2 studies), HbA_1c reduction from baseline ≥0.5% (n = 1 study), HbA_1c reduction from baseline ≥1% (n = 1 study), and HbA_1c reduction from baseline ≥1% or HbA_1c <7.0% (n = 1 study) (Supplementary Appendix 6). However, results from single studies showed that CGM, versus usual care/SMBG, increased the risk of HbA_1c reduction from baseline by ≥10% and proportion with HbA_1c level <8.0%: RR 1.53 (95% CI 1.07, 2.19) and 1.88 (1.30, 2.72), respectively (Supplementary Appendix 6).

In comparing CGM with usual care/SMBG, there were no differences in FPG (n = 3 studies) or mean glucose concentrations (n = 4 studies) but there was an increase in levels of minimum glucose (n = 2 studies) and maximum glucose (n = 2 studies) concentrations: MD −8.34 mg/dL (95% CI −29.33, 12.64), −0.09 mg/dL (−13.71, 13.53), 11.69 mg/dL (5.77, 17.60), and 18.17 mg/dL (8.83, 27.51), respectively (Supplementary Appendix 7).

isCGM, versus usual care/SMBG, reduced mean glucose (n = 4 studies) and FPG levels (n = 1 study): MD −7.52 mg/dL (95% CI −14.29, −0.75) and −18.00 mg/dL (−31.99, −4.01), respectively (Supplementary Appendix 8).

In comparing CGM with usual care/SMBG, there was no difference in the % TIR of 70–180 mg/dL (n = 6 studies), whereas investigators of a single study reported an increased time/day in range of 70–180 mg/dL: MD 2.51% (95% CI −6.08, 11.09) and 46.00 min (23.18, 68.82), respectively (Supplementary Appendix 9).

isCGM, versus usual care/SMBG, increased % TIR of 70–180 mg/dL (n = 2 studies) and time/day with glucose range of 70–180 mg/dL (n = 3 studies): MD 8.94% (95% CI 4.07, 13.82) and 1.11 h (0.13, 2.09), respectively (Supplementary Appendix 9).

In comparing CGM with usual care/SMBG, pooled results showed no differences in some measures of TBR including % time <54 mg/dL (n = 2 studies), % time <70 mg/dL (n = 5 studies), and time/day in glucose range <50 mg/dL (n = 2 studies): MD −0.20% (95% CI −0.65, 0.25), −0.36% (−0.78, 0.05), and −5.60 min (−21.26, 10.06), respectively (Supplementary Appendix 10). Single reports showed reductions in % time <60 mg/dL, time/day in glucose range <60 mg/dL, and time/day in glucose range <70 mg/dL (Supplementary Appendix 10).

In comparing isCGM with usual care/SMBG, there were no differences in several measures of TBR (Supplementary Appendix 11).

In comparing CGM with usual care/SMBG, pooled results showed an increased % time >140 mg/dL (n = 2 studies) and reduced % time >180 mg/dL (n = 4 studies), with no difference observed for % time >250 mg/dL (n = 3 studies): MD 10.43% (95% CI 3.69, 17.18), −7.70% (−14.70, −0.70), and −6.60% (−16.43, 3.23), respectively (Supplementary Appendix 12). Single reports showed a reduction in % time >300 mg/dL and time/day in glucose range >250 mg/dL, an increase in time/day in glucose range >300 mg/dL, and no difference in time/day in glucose range >180 mg/dL (Supplementary Appendix 12).

In comparing isCGM with usual care/SMBG, pooled results showed no difference in time/day with glucose >180 mg/dL (n = 3 studies) and a reduction in time/day with glucose >240 mg/dL % (n = 3 studies): MD −1.10 h (95% CI −2.45, 0.24) and −0.91 h (−1.52, −0.31), respectively (Supplementary Appendix 13). Single reports showed a reduction in % time >180 mg/dL and time/day with glucose >300 mg/dL (Supplementary Appendix 13).

In comparing CGM with usual care/SMBG, there were no differences in various measures of glucose variability such as the coefficient of variation (n = 3 studies), SD of mean glucose (n = 2 studies), amplitude of glucose excursions (n = 1 study), M index (n = 1 study), and number of hyperglycemic excursions (n = 1 study): MD −0.40 (95% CI −2.79, 2.00), 0.09 (−7.74, 7.92), −7.00 (−30.91, 16.91), 10.00 (−20.11, 40.11), and −0.70 (−1.92, 0.52), respectively (Supplementary Appendix 14).

The continuous overlapping net glycemic action (CONGA) metric determines the variance (SD) between a current blood glucose reading and one obtained (n) hours ago. In comparing isCGM with usual care/SMBG, pooled analysis showed a reduction in SD of mean glucose (n = 4 studies), CONGA 2 h (n = 3 studies), and CONGA 4 h (n = 2 studies), with no differences in coefficient of variation (n = 4 studies), low blood glucose index (n = 2 studies), blood glucose risk index (n = 2 studies), and mean amplitude of glycemic excursions (n = 2 studies): MD −3.62 mg/dL (95% CI −5.95, −1.29), −3.61 mg/dL (−6.14, −1.07), −5.83 mg/dL (−9.78, −1.88), −0.97% (−2.23, 0.28), −0.09 (−0.60, 0.43), −1.03 (−2.22, 0.15), and −10.47 mg/dL (−21.05, 0.12), respectively (Supplementary Appendix 15). Single reports showed no differences in high blood glucose index, mean of daily difference, CONGA 1 h, or SD of glucose rate of change but a reduction in CONGA 6 h (Supplementary Appendix 15).

When CGM and usual care/SMBG were compared, there were no differences in weight (n = 4 studies), BMI (n = 5 studies), or waist circumference (n = 1 study): MD −1.32 kg (95% CI −5.99, 3.35), 0.30 kg/m² (−1.18, 1.79), and −4.60 cm (−10.13, 0.93), respectively (Supplementary Appendix 16). In comparing CGM with usual care/SMBG, single study reports showed no differences in various lifestyle factors such as total calorie intake, cholesterol intake, and physical activity (Supplementary Appendix 17).

In comparing isCGM with usual care/SMBG, single study reports showed no differences in weight or BMI but a reduction in waist circumference (Supplementary Appendix 18).

In comparing CGM with usual care/SMBG, there were no differences in systolic blood pressure (n = 5 studies) or diastolic blood pressure (n = 5 studies): MD −1.10 mmHg (95% CI −6.63, 4.43) and −0.37 mmHg (−2.74, 1.99), respectively (Supplementary Appendix 19).

In comparing isCGM with usual care/SMBG, single study reports showed an increase in systolic blood pressure but no difference in diastolic blood pressure (Supplementary Appendix 20).

In comparing CGM with usual care/SMBG, there were no differences in levels of total cholesterol (n = 3 studies), HDL cholesterol (n = 3 studies), LDL cholesterol (n = 3 studies), or triglycerides (n = 3 studies): MD 0.16 mmol/L (95% CI −0.41, 0.73), 0.03 mmol/L (−0.20, 0.27), 0.18 mmol/L (−0.13, 0.48), and −0.22 mmol/L (−0.62, 0.18), respectively (Supplementary Appendix 21).

In comparing isCGM with usual care/SMBG, single study reports showed no differences in levels of total cholesterol, HDL cholesterol, LDL cholesterol, or triglycerides (Supplementary Appendix 22).

Pooled analysis of two studies showed that CGM achieved a significant reduction in glycemic medications compared with usual care/SMBG: MD −0.67 (95% CI −1.20, −0.13) (Supplementary Appendix 23).

CGM versus usual care/SMBG increased the risk of any adverse events (n = 5 studies) (e.g., sensor insertion site symptoms, infection, hypoglycemia, headache, and gastrointestinal symptoms), RR 1.22 (95% CI 1.01, 1.47), with no difference in the risk of specific adverse events such as hypoglycemia, severe hypoglycemia, and diabetic ketoacidosis, reported for single studies (Supplementary Appendix 24).

isCGM versus usual care/SMBG increased the risk of any adverse events (n = 3 studies) and device-related adverse events (n = 4 studies)—RR 1.30 (95% CI 1.05, 1.62) and 4.24 (1.80, 10.03), respectively—with no differences in the risk of serious adverse events, any hypoglycemia, severe hypoglycemia, severe hyperglycemia, or other measures of hypoglycemia and hyperglycemia (Supplementary Appendix 25).

In comparing CGM with usual care/SMBG, there were no differences in diabetes distress score (n = 3 studies), quality of life score (n = 2 studies), well-being score (n = 1 study), or depression score (n = 1 study): SMD −0.05 (95% CI −0.29, 0.19), 0.12 (−0.18, 0.41), −0.22 (−0.54, 0.10), and −0.16 (−0.92, 0.60), respectively (Supplementary Appendix 26). However, CGM, versus usual care/SMBG, reduced satisfaction scores (n = 3 studies): SMD −0.54 (−0.98, −0.11) (Supplementary Appendix 26).

isCGM, versus usual care/SMBG, increased satisfaction scores (n = 5 studies), with no differences in scores for diabetes distress (n = 2 studies), emotional well-being (n = 1 study), or quality of life (n = 1 study): SMD 0.44 (95% CI 0.29, 0.59), −0.02 (−0.21, 0.18), 0.05 (−0.18, 0.27), and −0.10 (−0.34, 0.15), respectively (Supplementary Appendix 27).

The funnel plot for change in HbA_1c was symmetrical on visual inspection, implying little evidence of small study effects or publication bias (Supplementary Appendix 28). This was consistent with Egger regression test (P value of 0.53).

GRADE certainty of the evidence ranged from moderate to very low (Supplementary Appendix 29).

Our findings revealed that both CGM and isCGM were associated with significant reductions in HbA_1c compared with usual care/SMBG, though the reduction was greater for isCGM (0.31% vs. 0.19%). The effect of CGM on HbA_1c levels was not modified by several clinically relevant characteristics. CGM was associated with a reduction in the need for glycemic medication escalation. isCGM but not CGM had a favorable effect on TIR metrics. Evaluation of other outcomes showed that CGM did not significantly impact glucose concentrations, glucose variability, measures of body composition, blood pressure, or lipid levels. However, isCGM reduced glucose concentrations and some measures of glucose variability. On the safety front, both CGM and isCGM increased the risk of any adverse events with no impact on hypoglycemia, but isCGM specifically increased the risk of device-related adverse events. Patient satisfaction scores were lower with CGM but higher with isCGM compared with usual care/SMBG. The certainty of the evidence varied from moderate to very low (Supplementary Appendix 4). Though a number of previous related reviews have been conducted on the topic, they are not similar in scope to the current review and therefore cannot be directly compared. Nevertheless, given some similarities in outcomes, they deserve mention and discussion. Ida et al. (11) in 2019 evaluated the impact of CGM on HbA_1c levels, weight, blood pressure, and frequency of hypoglycemia in patients with T2D based on a pooled analysis of seven RCTs. Their results showed that HbA_1c levels were significantly lower in the CGM group than in the SMBG group (11). Consistent with our findings, no differences in body weight or blood pressure were observed between the CGM and SMBG groups. Furthermore, a subgroup analysis by type of CGM (real time and retrospective), on HbA_1c levels, suggested that real-time CGM might be more effective than retrospective CGM, but there was no strong evidence of effect modification (11). Janapala et al. (23), in another review, published in 2019, which was based on a pooled analysis of five RCTs and only included evaluation of change in HbA_1c, reported a 0.25% mean reduction in HbA_1c levels in comparing CGM with SMBG. Dicembrini et al. (24) also in 2019 assessed the effect of continuous subcutaneous insulin infusion, CGM, and the combination of the two on glycemic control in patients with T2D. Pooled analysis of five RCTs for the comparison of CGM and SMBG showed a mean reduction in HbA_1c levels of 0.28% (24). In a number of reviews investigators have evaluated the efficacy and safety of isCGM in comparison with SMBG, but these were conducted in both populations with T1D and populations with T2D and fewer than three studies contributed to the pooled analysis for T2D (25,26).

Unlike in prior reviews, we assessed CGM and isCGM with diverse outcome measures from 26 trials. Our synthesis reveals gaps in evidence for T2D patients. Few large definitive trials exist, and most have short durations (8.0–34.8 weeks), with only two isCGM trials lasting 52 weeks. Long-term effects remain uncertain.

Health care providers should consider the glycemic advantages and reduction of medications with CGM and isCGM use. However, these glucose monitoring systems did not show significant impact on body weight, lipid profiles, blood pressure, or lifestyle changes, indicating that these aspects should continue to be addressed independently of their use, with the caveat that studies were relatively short and longer use may have an effect on these parameters. However, health care providers must remain vigilant about the increased risk of any adverse events associated with the use of CGM and isCGM, with a particular focus on isCGM given the increased risk of device-related adverse events. It is however, comforting to know that robust data based on real-world evidence suggests that the use of isCGM significantly lowers the incidence of diabetes complications such as diabetic ketoacidosis, severe hypoglycemia, and hyperglycemia (27). In a recent review of 10 observational and interventional studies where investigators sought to evaluate the efficacy of both CGM and isCGM for the management of diabetes in primary care settings, both monitoring systems were observed to be effective at reducing HbA_1c levels compared with usual care (28). The authors acknowledged that most of the studies had substantial biases and were of short-term duration (28). Our study results showed that patient satisfaction scores were higher with isCGM but relatively lower with CGM compared with usual care/SMBG, emphasizing the importance of considering patient preferences and, crucially, providing adequate support and education when recommending CGM. The lower satisfaction score for CGM was unexpected, but this may reflect the fact that for a number of the studies contributing to the pooled analysis (two of three) recruitment started more than a decade ago (2010 and 2012); these were periods when the CGM systems were less user-friendly. Over the years, these monitoring systems have undergone improvements and are more user-friendly. Overall, CGM and isCGM emerge as potential valuable tools in improving glycemic control and reducing HbA_1c levels in patients with T2D.

In this systematic review and meta-analysis we comprehensively assess CGM and isCGM benefits and harms in T2D patients. However, limitations include small sample sizes, single-study outcomes, and heterogeneity due to population variations, interventions, and outcome definitions. Generalizability to younger adults is uncertain, as most studies involved patients aged 53–70 years.

Specific studies are needed in younger adults given the increasing prevalence of T2D in this population group. Except for the change in HbA_1c outcome, all other end points were based on pooled analyses of <10 studies, which limited the extent of comprehensive evaluation such as subgroup analysis by relevant clinical characteristics. Finally, these findings need to be interpreted in light of the fact that the publication dates of these trials spanned 2008–2023 and that these monitoring systems have undergone technological improvements over the years.

A reduction in HbA_1c levels was demonstrated with both CGM and isCGM use in individuals with T2D, and unlike with CGM, isCGM use was associated with improved user satisfaction The impact of these devices on body composition, blood pressure, and lipid levels remains unclear, while both CGM and isCGM use were associated with increased risk of adverse events. Further research is necessary to gain a deeper understanding of the long-term clinical implications of the use of CGM and isCGM in patients with T2D.

Funding. P.C., S.K.K., and S.S. are supported by the National Institute for Health Research, Applied Research Collaboration East Midlands, and the National Institute for Health Research Leicester Biomedical Research Centre.

Duality of Interest. S.S. reports personal fees from Amgen, AstraZeneca, Napp Pharmaceuticals Limited (NAPP), Eli Lilly, Merck Sharp & Dohme, Novartis, Novo Nordisk, Roche, Sanofi, Abbott, and Boehringer Ingelheim. P.C. has received personal fees from Abbott Diabetes Care, Dexcom, Diasend, Eli Lilly, Insulet, Medtronic, Novo Nordisk, Roche, and Sanofi. A.A.R. has received honoraria from Abbott Diabetes Care but not in relation to this work. No other potential conflicts of interest relevant to this article were reported.

The authors received no industry financial support for the research, authorship, or publication of this article.

Author Contributions. S.S. contributed to study design, literature search, data collection, data interpretation, and writing. S.K.K. contributed to study design, literature search, data collection, analysis, data interpretation, and writing. A.A.R. contributed to study design, data interpretation, and writing. P.C. contributed to study design, data interpretation, and writing.